zompist bboard

Sihler says that in a sequence of two resonants between non-resonants, the second is always syllabic. This means that, say, dwrt is realised as /dwr.t/ rather than /durt/, which strikes me as strange since [r] is less vocalic than [w] and so /durt/ would be more natural. Where's the flaw?

bricka wrote:Sihler says that in a sequence of two resonants between non-resonants, the second is always syllabic. This means that, say, dwrt is realised as /dwr.t/ rather than /durt/, which strikes me as strange since [r] is less vocalic than [w] and so /durt/ would be more natural. Where's the flaw?

I usually trust Sihler, so give me chapter & verse so that I can see what he actually says in context.

I presume that this is about PIE, and not other languages. The rule clearly doesn't apply in Latin sunt, for example.

If the rule applies in PIE, it can only apply in my opinion in the worldview of people so enamoured of underlying forms that all common sense departs. Some such people will claim that PIE had no *i and *u, and any forms that seem to have had such vowels on the surface could be analysed phonemically as /CyC/ and /CwC/, respectively, with C standing for any non-resonant.

Personally, I see nothing wrong with *nisdos 'nest', and would not see an analysis as *nysdos an improvement. If your hypothetical *durt ever came up, I would embrace it with all my heart! !

Ringe describes PIE as having both *i *u and *y *w, and proposes that *y *w *n *m *r *l and the laryngeals are underlyingly nonsyllabic.

Thus, they are syllabified from right to left: if the rightmost in a sequence is adjacent to a syllabic (initially, only a vowel) it remains nonsyllabic. Otherwise, it becomes syllabic, and we move on to check the next segment.

There are a few exceptions to this, in that it does not apply to certain morphemes: the infix *n in some Present verb stems will never syllabify.

bricka wrote:/durt/ would be more natural.

This isn't phonetic truth, this is phonological/dependent on language. It doesn't matter if [r] is less sonorant (and if /r/ was actually [ɹ] or something else, the situation changes entirely) – the point is it forms a phonological class with /w/, maybe as a "continuant" rather than "resonant" or "sonorant" so that you can see the similarity... You're also confused because [w] has a vowel equivalent, which [r] doesn't.

I know nothing of PIE in particular, but sometimes you just have to throw out these little prejudices, in linguistics in general....

gsandi wrote:I usually trust Sihler, so give me chapter & verse so that I can see what he actually says in context.

The relevant passage is on page 91 of his New Comparative Grammar of Greek and Latin; it's a commentary on item 2 in section 93: a resonant is syllabic "between a resonant and a following obstruent, laryngeal, or word boundary". So in dwrt, and also in dwr and dawr, the /r/ is syllabic.

I think finlay has a point.

bricka wrote:

gsandi wrote:I usually trust Sihler, so give me chapter & verse so that I can see what he actually says in context.

The relevant passage is on page 91 of his New Comparative Grammar of Greek and Latin; it's a commentary on item 2 in section 93: a resonant is syllabic "between a resonant and a following obstruent, laryngeal, or word boundary". So in dwrt, and also in dwr and dawr, the /r/ is syllabic.

I think finlay has a point.

But Sihler himself gives two types of exceptions (e.g. *owim, *yung-), so that we have at best a pragmatic generalization, not an absolute rule.

Wikipedia at http://en.wikipedia.org/wiki/Sanskrit#Phonology says that Sanskrit 'r' was /r/, a trill. I had an old book that said it was the American /r\/, which seems to make more sense to me, but they may have just been making it easier for English speakers to understand. Still the idea of having a length distinction between long and short syllabic [r] seems difficult for me to accept. Does anyone know the answer? And I have the same question for PIE, which is why Im posting it here. The majoruity of early IE languages do seem to have something close to [r].

Also: There dont seem to be many examples of a stop followed by a laryngeal, do there? All I can find is when an ablaut vowel becomes zero-grade, and a few things like phter for father which may have started out as peH or something + -ter anyway . Yet sonorant + laryngeal is very common. Could it be that the aspirated stops arose from some previous change involving laryngeals that occurred next to (previously unaspirated) stops? Perhaps not, because aspirated consonants are very common indeed in PIE, too much to be accounted for by a change like that, but it makes me curious nonetheless.

Soap wrote:Wikipedia at http://en.wikipedia.org/wiki/Sanskrit#Phonology says that Sanskrit 'r' was /r/, a trill.

Where?

Still the idea of having a length distinction between long and short syllabic [r] seems difficult for me to accept.

"Long syllabic r (ṝ) is also quite marginal, occurring in the genitive plural of r-stems (e.g. mātṛ "mother" and pitṛ "father" have gen.pl. mātṝṇām and pitṝṇām)."

Does anyone know the answer?

I'm pretty sure it's a retroflex approximant. Damned if I can come up with a source for that, but a lot of langs in that general area have retroflex approximants, and that makes more sense with the phonology anyway.

Soap wrote:Still the idea of having a length distinction between long and short syllabic [r] seems difficult for me to accept.

Doesn't Slovak have that?

Soap wrote:Still the idea of having a length distinction between long and short syllabic [r] seems difficult for me to accept.

Serbo-Croatian has it.

Soap wrote:Still the idea of having a length distinction between long and short syllabic [r] seems difficult for me to accept.[r].

Dialects of English that have replaced historical phonemic vowel length with (often at least partly frozen) allophonic vowel length have that...

I have A Sanskrit Primer by Edward Delavan Perry from 1936, which says that "the r is clearly a lingual sound. It thus resembles the English smooth r, and like this seems to have been untrilled."

Cool. Wikipedia's wrong, then. Feel free to change it and add a link to the primer with page number, or let me know the page and I'll change it myself.